> OK. I will post some pictures of the raytracing results from super
> ellipsoids with e or n > 3. I also think generating multiple surfaces is a
> way to get sharp corners, although it's a little bit more complicated than a
> single surface.
Interesting images. Yeah, that's probably a multiple-surface
situation - individual NURBS surfaces with "kinks" in them generally
tend to cause problems, from what I can see...
> I found it's really view dependent. That's quite strange. So, what is this
> primitive usually used to do?
IIRC, cline was used originally in some analytical situations - it's
not a common primitive, and to the best of my knowledge we support it
only for legacy analytical models and not for any modern stuff
(certainly not for "normal" CSG modeling.) There are three options:
1. Document why it can't be expressed as a Brep and leave it at that.
2. Add a utility command to take a "cline" primitive and translate it
into it's "closest match" pipe primitive. That would a) allow people
who don't want the quirky cline behavior to get a pipe that behave
more like what they want, and b) provide a route to a Brep through the
pipe conversion routines. We eventually will want to be able to go
from cline to implicit pipe anyway, so it addresses two issues at
3. Pick a particular interpretation of the cline inputs for NURBS
creation and generate that Brep.
Personally, I'd suggest option two as the most useful feature even
though it's not directly NURBS related - then if you want brep to do
something intelligent with a cline, you could have cline's brep
routine call the cline->pipe routine first and then pipe->brep using
> As far as I'm concerned, that will be a great challenge for me. NURBS
> intersection is another GSOC project and is marked as 'hard'. So I will read
> the papers you offer in a few days, and find out whether I can manage to do
Sounds great! If, once you've read over the papers, you decide you
want to have a go at it we'll need to identify the approach and the
ordering for intermediate pieces to work on - there will be a few of
them. If you opt not to tackle it don't worry - there's still plenty
of other things to do. I mainly suggested this one because it's more
or less the next logical continuation of the conversion - essentially,
fully supporting the "comb" conversion to brep. Also, despite being
hard (or maybe because of it ;-) it is a fun problem for folks who
like this kind of thing.